]>
Commit | Line | Data |
---|---|---|
a4057fc9 | 1 | #!/usr/bin/env python |
2 | ||
3 | ## @package thtml2doxy_clang | |
4 | # Translates THtml C++ comments to Doxygen using libclang as parser. | |
5 | # | |
6 | # This code relies on Python bindings for libclang: libclang's interface is pretty unstable, and | |
7 | # its Python bindings are unstable as well. | |
8 | # | |
9 | # AST (Abstract Source Tree) traversal is performed entirely using libclang used as a C++ parser, | |
10 | # instead of attempting to write a parser ourselves. | |
11 | # | |
12 | # This code (expecially AST traversal) was inspired by: | |
13 | # | |
14 | # - [Implementing a code generator with libclang](http://szelei.me/code-generator/) | |
15 | # (this refers to API calls used here) | |
16 | # - [Parsing C++ in Python with Clang](http://eli.thegreenplace.net/2011/07/03/parsing-c-in-python-with-clang) | |
17 | # (outdated, API calls described there do not work anymore, but useful to understand some basic | |
18 | # concepts) | |
19 | # | |
20 | # Usage: | |
21 | # | |
6a98a5fd | 22 | # `thtml2doxy_clang [--stdout|-o] [-d] [--debug=DEBUG_LEVEL] file1 [file2 [file3...]]` |
a4057fc9 | 23 | # |
6a98a5fd | 24 | # Parameters: |
25 | # | |
26 | # - `--stdout|-o`: output all on standard output instead of writing files in place | |
27 | # - `-d`: enable debug mode (very verbose output) | |
28 | # - `--debug=DEBUG_LEVEL`: set debug level to one of `DEBUG`, `INFO`, `WARNING`, `ERROR`, | |
29 | # `CRITICAL` | |
30 | # | |
31 | # @author Dario Berzano, CERN | |
a4057fc9 | 32 | # @date 2014-12-05 |
33 | ||
34 | ||
35 | import sys | |
36 | import os | |
37 | import re | |
62671ba0 | 38 | import logging |
39 | import getopt | |
a4057fc9 | 40 | import clang.cindex |
41 | ||
42 | ||
43 | ## Brain-dead color output for terminal. | |
44 | class Colt(str): | |
45 | ||
46 | def red(self): | |
47 | return self.color('\033[31m') | |
48 | ||
49 | def green(self): | |
50 | return self.color('\033[32m') | |
51 | ||
52 | def yellow(self): | |
53 | return self.color('\033[33m') | |
54 | ||
55 | def blue(self): | |
56 | return self.color('\033[34m') | |
57 | ||
58 | def magenta(self): | |
59 | return self.color('\033[35m') | |
60 | ||
61 | def cyan(self): | |
62 | return self.color('\033[36m') | |
63 | ||
64 | def color(self, c): | |
65 | return c + self + '\033[m' | |
66 | ||
67 | ||
10f6b9f8 | 68 | ## Comment. |
69 | class Comment: | |
70 | ||
72604c43 | 71 | def __init__(self, lines, first_line, first_col, last_line, last_col, indent, func): |
10f6b9f8 | 72 | self.lines = lines |
73 | self.first_line = first_line | |
9220af47 | 74 | self.first_col = first_col |
10f6b9f8 | 75 | self.last_line = last_line |
9220af47 | 76 | self.last_col = last_col |
72604c43 | 77 | self.indent = indent |
10f6b9f8 | 78 | self.func = func |
79 | ||
72604c43 | 80 | def has_comment(self, line): |
81 | return line >= self.first_line and line <= self.last_line | |
82 | ||
10f6b9f8 | 83 | def __str__(self): |
9220af47 | 84 | return "<Comment for %s: [%d,%d:%d,%d] %s>" % (self.func, self.first_line, self.first_col, self.last_line, self.last_col, self.lines) |
10f6b9f8 | 85 | |
86 | ||
3018d1db | 87 | ## A data member comment. |
88 | class MemberComment: | |
89 | ||
90 | def __init__(self, text, is_transient, array_size, first_line, first_col, func): | |
91 | self.lines = [ text ] | |
92 | self.is_transient = is_transient | |
93 | self.array_size = array_size | |
94 | self.first_line = first_line | |
95 | self.first_col = first_col | |
96 | self.func = func | |
97 | ||
98 | def has_comment(self, line): | |
99 | return line == self.first_line | |
100 | ||
101 | def __str__(self): | |
102 | ||
103 | if self.is_transient: | |
104 | tt = '!transient! ' | |
105 | else: | |
106 | tt = '' | |
107 | ||
108 | if self.array_size is not None: | |
109 | ars = '[%s] ' % self.array_size | |
110 | else: | |
111 | ars = '' | |
112 | ||
113 | return "<MemberComment for %s: [%d,%d] %s%s%s>" % (self.func, self.first_line, self.first_col, tt, ars, self.lines[0]) | |
114 | ||
115 | ||
3b095296 | 116 | ## A dummy comment that removes comment lines. |
117 | class RemoveComment(Comment): | |
118 | ||
119 | def __init__(self, first_line, last_line): | |
120 | self.first_line = first_line | |
121 | self.last_line = last_line | |
122 | self.func = '<remove>' | |
123 | ||
124 | def __str__(self): | |
125 | return "<RemoveComment: [%d,%d]>" % (self.first_line, self.last_line) | |
126 | ||
127 | ||
a7136c98 | 128 | ## Parses method comments. |
a4057fc9 | 129 | # |
a7136c98 | 130 | # @param cursor Current libclang parser cursor |
131 | # @param comments Array of comments: new ones will be appended there | |
132 | def comment_method(cursor, comments): | |
133 | ||
134 | # we are looking for the following structure: method -> compound statement -> comment, i.e. we | |
135 | # need to extract the first comment in the compound statement composing the method | |
136 | ||
137 | in_compound_stmt = False | |
138 | expect_comment = False | |
139 | emit_comment = False | |
140 | ||
141 | comment = [] | |
142 | comment_function = cursor.spelling or cursor.displayname | |
143 | comment_line_start = -1 | |
144 | comment_line_end = -1 | |
145 | comment_col_start = -1 | |
146 | comment_col_end = -1 | |
147 | comment_indent = -1 | |
148 | ||
149 | for token in cursor.get_tokens(): | |
150 | ||
151 | if token.cursor.kind == clang.cindex.CursorKind.COMPOUND_STMT: | |
152 | if not in_compound_stmt: | |
153 | in_compound_stmt = True | |
154 | expect_comment = True | |
155 | comment_line_end = -1 | |
156 | else: | |
157 | if in_compound_stmt: | |
158 | in_compound_stmt = False | |
159 | emit_comment = True | |
a4057fc9 | 160 | |
a7136c98 | 161 | # tkind = str(token.kind)[str(token.kind).index('.')+1:] |
162 | # ckind = str(token.cursor.kind)[str(token.cursor.kind).index('.')+1:] | |
a4057fc9 | 163 | |
a7136c98 | 164 | if in_compound_stmt: |
a4057fc9 | 165 | |
a7136c98 | 166 | if expect_comment: |
a4057fc9 | 167 | |
a7136c98 | 168 | extent = token.extent |
169 | line_start = extent.start.line | |
170 | line_end = extent.end.line | |
a4057fc9 | 171 | |
a7136c98 | 172 | if token.kind == clang.cindex.TokenKind.PUNCTUATION and token.spelling == '{': |
173 | pass | |
a4057fc9 | 174 | |
a7136c98 | 175 | elif token.kind == clang.cindex.TokenKind.COMMENT and (comment_line_end == -1 or (line_start == comment_line_end+1 and line_end-line_start == 0)): |
176 | comment_line_end = line_end | |
177 | comment_col_end = extent.end.column | |
a4057fc9 | 178 | |
a7136c98 | 179 | if comment_indent == -1 or (extent.start.column-1) < comment_indent: |
180 | comment_indent = extent.start.column-1 | |
a4057fc9 | 181 | |
a7136c98 | 182 | if comment_line_start == -1: |
183 | comment_line_start = line_start | |
184 | comment_col_start = extent.start.column | |
185 | comment.extend( token.spelling.split('\n') ) | |
186 | ||
187 | # multiline comments are parsed in one go, therefore don't expect subsequent comments | |
188 | if line_end - line_start > 0: | |
189 | emit_comment = True | |
190 | expect_comment = False | |
a4057fc9 | 191 | |
a7136c98 | 192 | else: |
193 | emit_comment = True | |
194 | expect_comment = False | |
9a13b5a2 | 195 | |
a7136c98 | 196 | if emit_comment: |
9a13b5a2 | 197 | |
a7136c98 | 198 | comment = refactor_comment( comment ) |
4e465d49 | 199 | |
a7136c98 | 200 | if len(comment) > 0: |
201 | logging.debug("Comment found for function %s" % Colt(comment_function).magenta()) | |
202 | comments.append( Comment(comment, comment_line_start, comment_col_start, comment_line_end, comment_col_end, comment_indent, comment_function) ) | |
72604c43 | 203 | |
a7136c98 | 204 | comment = [] |
205 | comment_line_start = -1 | |
206 | comment_line_end = -1 | |
207 | comment_col_start = -1 | |
208 | comment_col_end = -1 | |
209 | comment_indent = -1 | |
a4057fc9 | 210 | |
a7136c98 | 211 | emit_comment = False |
212 | break | |
78aaad66 | 213 | |
a4057fc9 | 214 | |
3018d1db | 215 | ## Parses comments to class data members. |
216 | # | |
217 | # @param cursor Current libclang parser cursor | |
218 | # @param comments Array of comments: new ones will be appended there | |
219 | def comment_datamember(cursor, comments): | |
220 | ||
221 | # Note: libclang 3.5 seems to have problems parsing a certain type of FIELD_DECL, so we revert | |
222 | # to a partial manual parsing. When parsing fails, the cursor's "extent" is not set properly, | |
223 | # returning a line range 0-0. We therefore make the not-so-absurd assumption that the datamember | |
224 | # definition is fully on one line, and we take the line number from cursor.location. | |
225 | ||
226 | line_num = cursor.location.line | |
227 | raw = None | |
3b095296 | 228 | prev = None |
229 | found = False | |
3018d1db | 230 | |
231 | # Huge overkill | |
232 | with open(str(cursor.location.file)) as fp: | |
233 | cur_line = 0 | |
234 | for raw in fp: | |
235 | cur_line = cur_line + 1 | |
236 | if cur_line == line_num: | |
3b095296 | 237 | found = True |
3018d1db | 238 | break |
3b095296 | 239 | prev = raw |
240 | ||
241 | assert found, 'A line that should exist was not found in file' % cursor.location.file | |
3018d1db | 242 | |
3e14e47f | 243 | recomm = r'(//(!)|///?)(\[(.*?)\])?<?\s*(.*?)\s*$' |
3b095296 | 244 | recomm_doxyary = r'^\s*///\s*(.*?)\s*$' |
3018d1db | 245 | |
246 | mcomm = re.search(recomm, raw) | |
247 | if mcomm: | |
248 | member_name = cursor.spelling; | |
249 | is_transient = mcomm.group(2) is not None | |
250 | array_size = mcomm.group(4) | |
251 | text = mcomm.group(5) | |
252 | ||
253 | col_num = mcomm.start()+1; | |
254 | ||
3b095296 | 255 | if array_size is not None and prev is not None: |
256 | # ROOT arrays with comments already converted to Doxygen have the member description on the | |
257 | # previous line | |
258 | mcomm_doxyary = re.search(recomm_doxyary, prev) | |
259 | if mcomm_doxyary: | |
260 | text = mcomm_doxyary.group(1) | |
261 | comments.append(RemoveComment(line_num-1, line_num-1)) | |
3018d1db | 262 | |
263 | logging.debug('Comment found for member %s' % Colt(member_name).magenta()) | |
264 | ||
265 | comments.append( MemberComment( | |
266 | text, | |
267 | is_transient, | |
268 | array_size, | |
269 | line_num, | |
270 | col_num, | |
271 | member_name )) | |
272 | ||
3b095296 | 273 | else: |
274 | assert False, 'Regular expression does not match member comment' | |
275 | ||
3018d1db | 276 | |
c7ea28e6 | 277 | ## Parses class description (beginning of file). |
278 | # | |
279 | # The clang parser does not work in this case so we do it manually, but it is very simple: we keep | |
280 | # the first consecutive sequence of single-line comments (//) we find - provided that it occurs | |
281 | # before any other comment found so far in the file (the comments array is inspected to ensure | |
282 | # this). | |
283 | # | |
284 | # Multi-line comments (/* ... */) are not considered as they are commonly used to display | |
285 | # copyright notice. | |
286 | # | |
287 | # @param filename Name of the current file | |
288 | # @param comments Array of comments: new ones will be appended there | |
289 | def comment_classdesc(filename, comments): | |
290 | ||
291 | recomm = r'^\s*///?\s*(.*?)\s*/*\s*$' | |
a00f7da1 | 292 | |
acd85b9a | 293 | reclass_doxy = r'(?i)^\\class:?\s*(.*?)\s*$' |
294 | class_name_doxy = None | |
a00f7da1 | 295 | |
296 | reauthor = r'(?i)\\?authors?:?\s*(.*?)\s*(,?\s*([0-9./-]+))?\s*$' | |
297 | redate = r'(?i)\\?date:?\s*([0-9./-]+)\s*$' | |
298 | author = None | |
299 | date = None | |
300 | ||
c7ea28e6 | 301 | comment_lines = [] |
302 | ||
303 | start_line = -1 | |
304 | end_line = -1 | |
305 | ||
306 | line_num = 0 | |
307 | ||
308 | with open(filename, 'r') as fp: | |
309 | ||
310 | for raw in fp: | |
311 | ||
312 | line_num = line_num + 1 | |
313 | ||
314 | if raw.strip() == '': | |
315 | # Skip empty lines | |
316 | end_line = line_num - 1 | |
317 | continue | |
318 | ||
319 | mcomm = re.search(recomm, raw) | |
320 | if mcomm: | |
321 | ||
acd85b9a | 322 | if start_line == -1 and len(comment_lines) == 0: |
c7ea28e6 | 323 | |
acd85b9a | 324 | # First line. Check that we do not overlap with other comments |
c7ea28e6 | 325 | comment_overlaps = False |
326 | for c in comments: | |
327 | if c.has_comment(line_num): | |
328 | comment_overlaps = True | |
329 | break | |
330 | ||
331 | if comment_overlaps: | |
332 | # No need to look for other comments | |
333 | break | |
334 | ||
335 | start_line = line_num | |
336 | ||
a00f7da1 | 337 | append = True |
338 | ||
acd85b9a | 339 | mclass_doxy = re.search(reclass_doxy, mcomm.group(1)) |
340 | if mclass_doxy: | |
341 | class_name_doxy = mclass_doxy.group(1) | |
a00f7da1 | 342 | append = False |
acd85b9a | 343 | else: |
a00f7da1 | 344 | mauthor = re.search(reauthor, mcomm.group(1)) |
345 | if mauthor: | |
346 | author = mauthor.group(1) | |
347 | if date is None: | |
348 | # Date specified in the standalone \date field has priority | |
349 | date = mauthor.group(2) | |
350 | append = False | |
351 | else: | |
352 | mdate = re.search(redate, mcomm.group(1)) | |
353 | if mdate: | |
354 | date = mdate.group(1) | |
355 | append = False | |
356 | ||
357 | if append: | |
acd85b9a | 358 | comment_lines.append( mcomm.group(1) ) |
c7ea28e6 | 359 | |
360 | else: | |
361 | if len(comment_lines) > 0: | |
362 | # End of our comment | |
363 | if end_line == -1: | |
364 | end_line = line_num - 1 | |
365 | break | |
366 | ||
acd85b9a | 367 | if class_name_doxy is None: |
c7ea28e6 | 368 | |
acd85b9a | 369 | # No \class specified: guess it from file name |
c7ea28e6 | 370 | reclass = r'^(.*/)?(.*?)(\..*)?$' |
371 | mclass = re.search( reclass, filename ) | |
372 | if mclass: | |
acd85b9a | 373 | class_name_doxy = mclass.group(2) |
c7ea28e6 | 374 | else: |
375 | assert False, 'Regexp unable to extract classname from file' | |
376 | ||
acd85b9a | 377 | # Prepend \class specifier (and an empty line) |
378 | comment_lines[:0] = [ '\\class ' + class_name_doxy ] | |
379 | ||
380 | # Append author and date if they exist | |
381 | comment_lines.append('') | |
382 | ||
383 | if author is not None: | |
384 | comment_lines.append( '\\author ' + author ) | |
385 | ||
386 | if date is not None: | |
387 | comment_lines.append( '\\date ' + date ) | |
388 | ||
389 | comment_lines = refactor_comment(comment_lines) | |
390 | logging.debug('Comment found for class %s' % Colt(class_name_doxy).magenta()) | |
391 | comments.append(Comment( | |
392 | comment_lines, | |
393 | start_line, 1, end_line, 1, | |
394 | 0, class_name_doxy | |
395 | )) | |
396 | ||
397 | ||
a7136c98 | 398 | ## Traverse the AST recursively starting from the current cursor. |
399 | # | |
400 | # @param cursor A Clang parser cursor | |
3bd2e2f0 | 401 | # @param filename Name of the current file |
a7136c98 | 402 | # @param comments Array of comments: new ones will be appended there |
403 | # @param recursion Current recursion depth | |
3bd2e2f0 | 404 | def traverse_ast(cursor, filename, comments, recursion=0): |
405 | ||
406 | # libclang traverses included files as well: we do not want this behavior | |
407 | if cursor.location.file is not None and str(cursor.location.file) != filename: | |
408 | logging.debug("Skipping processing of included %s" % cursor.location.file) | |
409 | return | |
533918c9 | 410 | |
a7136c98 | 411 | text = cursor.spelling or cursor.displayname |
412 | kind = str(cursor.kind)[str(cursor.kind).index('.')+1:] | |
533918c9 | 413 | |
a7136c98 | 414 | indent = '' |
415 | for i in range(0, recursion): | |
416 | indent = indent + ' ' | |
a4057fc9 | 417 | |
dc396a82 | 418 | if cursor.kind == clang.cindex.CursorKind.CXX_METHOD or cursor.kind == clang.cindex.CursorKind.CONSTRUCTOR or cursor.kind == clang.cindex.CursorKind.DESTRUCTOR: |
533918c9 | 419 | |
a7136c98 | 420 | # cursor ran into a C++ method |
1f2b1b91 | 421 | logging.debug( "%5d %s%s(%s)" % (cursor.location.line, indent, Colt(kind).magenta(), Colt(text).blue()) ) |
a7136c98 | 422 | comment_method(cursor, comments) |
a4057fc9 | 423 | |
3018d1db | 424 | elif cursor.kind == clang.cindex.CursorKind.FIELD_DECL: |
425 | ||
426 | # cursor ran into a data member declaration | |
1f2b1b91 | 427 | logging.debug( "%5d %s%s(%s)" % (cursor.location.line, indent, Colt(kind).magenta(), Colt(text).blue()) ) |
3018d1db | 428 | comment_datamember(cursor, comments) |
429 | ||
a4057fc9 | 430 | else: |
431 | ||
1f2b1b91 | 432 | logging.debug( "%5d %s%s(%s)" % (cursor.location.line, indent, kind, text) ) |
a4057fc9 | 433 | |
4e465d49 | 434 | for child_cursor in cursor.get_children(): |
3bd2e2f0 | 435 | traverse_ast(child_cursor, filename, comments, recursion+1) |
a4057fc9 | 436 | |
c7ea28e6 | 437 | if recursion == 0: |
438 | comment_classdesc(filename, comments) | |
439 | ||
533918c9 | 440 | |
4e465d49 | 441 | ## Remove garbage from comments and convert special tags from THtml to Doxygen. |
442 | # | |
9eb7bab8 | 443 | # @param comment An array containing the lines of the original comment |
4e465d49 | 444 | def refactor_comment(comment): |
a4057fc9 | 445 | |
9eb7bab8 | 446 | recomm = r'^(/{2,}|/\*)?\s*(.*?)\s*((/{2,})?\s*|\*/)$' |
b62a05f1 | 447 | regarbage = r'^[\s*=-_#]+$' |
9eb7bab8 | 448 | |
449 | new_comment = [] | |
450 | insert_blank = False | |
451 | wait_first_non_blank = True | |
452 | for line_comment in comment: | |
453 | mcomm = re.search( recomm, line_comment ) | |
454 | if mcomm: | |
455 | new_line_comment = mcomm.group(2) | |
b62a05f1 | 456 | mgarbage = re.search( regarbage, new_line_comment ) |
fc54cb81 | 457 | |
b62a05f1 | 458 | if new_line_comment == '' or mgarbage is not None: |
9eb7bab8 | 459 | insert_blank = True |
4e465d49 | 460 | else: |
9eb7bab8 | 461 | if insert_blank and not wait_first_non_blank: |
462 | new_comment.append('') | |
fc54cb81 | 463 | insert_blank = False |
9eb7bab8 | 464 | wait_first_non_blank = False |
465 | new_comment.append( new_line_comment ) | |
fc54cb81 | 466 | |
9eb7bab8 | 467 | else: |
468 | assert False, 'Comment regexp does not match' | |
4e465d49 | 469 | |
470 | return new_comment | |
a4057fc9 | 471 | |
472 | ||
72604c43 | 473 | ## Rewrites all comments from the given file handler. |
474 | # | |
475 | # @param fhin The file handler to read from | |
476 | # @param fhout The file handler to write to | |
477 | # @param comments Array of comments | |
478 | def rewrite_comments(fhin, fhout, comments): | |
479 | ||
480 | line_num = 0 | |
72604c43 | 481 | in_comment = False |
482 | skip_empty = False | |
3b095296 | 483 | comm = None |
484 | prev_comm = None | |
72604c43 | 485 | |
3018d1db | 486 | rindent = r'^(\s*)' |
487 | ||
72604c43 | 488 | for line in fhin: |
489 | ||
490 | line_num = line_num + 1 | |
491 | ||
3b095296 | 492 | # Find current comment |
493 | prev_comm = comm | |
494 | comm = None | |
495 | for c in comments: | |
496 | if c.has_comment(line_num): | |
497 | comm = c | |
498 | ||
499 | if comm: | |
72604c43 | 500 | |
3018d1db | 501 | if isinstance(comm, MemberComment): |
3b095296 | 502 | non_comment = line[ 0:comm.first_col-1 ] |
3018d1db | 503 | |
504 | if comm.array_size is not None: | |
505 | ||
506 | mindent = re.search(rindent, line) | |
507 | if comm.is_transient: | |
508 | tt = '!' | |
509 | else: | |
510 | tt = '' | |
511 | ||
512 | # Special case: we need multiple lines not to confuse ROOT's C++ parser | |
3b095296 | 513 | fhout.write('%s/// %s\n%s//%s[%s]\n' % ( |
3018d1db | 514 | mindent.group(1), |
515 | comm.lines[0], | |
516 | non_comment, | |
517 | tt, | |
518 | comm.array_size | |
519 | )) | |
520 | ||
521 | else: | |
522 | ||
523 | if comm.is_transient: | |
524 | tt = '!' | |
525 | else: | |
526 | tt = '/' | |
527 | ||
528 | fhout.write('%s//%s< %s\n' % ( | |
529 | non_comment, | |
530 | tt, | |
531 | comm.lines[0] | |
532 | )) | |
533 | ||
3b095296 | 534 | elif isinstance(comm, RemoveComment): |
535 | # Do nothing: just skip line | |
536 | pass | |
3018d1db | 537 | |
3b095296 | 538 | elif prev_comm is None: |
539 | # Beginning of a new comment block of type Comment | |
72604c43 | 540 | in_comment = True |
541 | ||
3b095296 | 542 | # Extract the non-comment part and print it if it exists |
543 | non_comment = line[ 0:comm.first_col-1 ].rstrip() | |
72604c43 | 544 | if non_comment != '': |
545 | fhout.write( non_comment + '\n' ) | |
546 | ||
547 | else: | |
3b095296 | 548 | |
72604c43 | 549 | if in_comment: |
550 | ||
3b095296 | 551 | # We have just exited a comment block of type Comment |
72604c43 | 552 | in_comment = False |
553 | ||
3b095296 | 554 | # Dump revamped comment, if applicable |
72604c43 | 555 | text_indent = '' |
3b095296 | 556 | for i in range(0,prev_comm.indent): |
72604c43 | 557 | text_indent = text_indent + ' ' |
558 | ||
3b095296 | 559 | for lc in prev_comm.lines: |
72604c43 | 560 | fhout.write( "%s/// %s\n" % (text_indent, lc) ); |
561 | fhout.write('\n') | |
562 | skip_empty = True | |
563 | ||
72604c43 | 564 | line_out = line.rstrip('\n') |
565 | if skip_empty: | |
566 | skip_empty = False | |
567 | if line_out.strip() != '': | |
568 | fhout.write( line_out + '\n' ) | |
569 | else: | |
570 | fhout.write( line_out + '\n' ) | |
571 | ||
572 | ||
a4057fc9 | 573 | ## The main function. |
574 | # | |
62671ba0 | 575 | # Return value is the executable's return value. |
a4057fc9 | 576 | def main(argv): |
577 | ||
62671ba0 | 578 | # Setup logging on stderr |
72604c43 | 579 | log_level = logging.INFO |
62671ba0 | 580 | logging.basicConfig( |
581 | level=log_level, | |
582 | format='%(levelname)-8s %(funcName)-20s %(message)s', | |
583 | stream=sys.stderr | |
584 | ) | |
585 | ||
586 | # Parse command-line options | |
43ee56ee | 587 | output_on_stdout = False |
62671ba0 | 588 | try: |
43ee56ee | 589 | opts, args = getopt.getopt( argv, 'od', [ 'debug=', 'stdout' ] ) |
62671ba0 | 590 | for o, a in opts: |
591 | if o == '--debug': | |
592 | log_level = getattr( logging, a.upper(), None ) | |
593 | if not isinstance(log_level, int): | |
594 | raise getopt.GetoptError('log level must be one of: DEBUG, INFO, WARNING, ERROR, CRITICAL') | |
595 | elif o == '-d': | |
596 | log_level = logging.DEBUG | |
43ee56ee | 597 | elif o == '-o' or o == '--stdout': |
598 | logging.debug('Output on stdout instead of replacing original files') | |
599 | output_on_stdout = True | |
62671ba0 | 600 | else: |
601 | assert False, 'Unhandled argument' | |
602 | except getopt.GetoptError as e: | |
603 | logging.fatal('Invalid arguments: %s' % e) | |
604 | return 1 | |
605 | ||
606 | logging.getLogger('').setLevel(log_level) | |
607 | ||
a4057fc9 | 608 | # Attempt to load libclang from a list of known locations |
609 | libclang_locations = [ | |
610 | '/usr/lib/llvm-3.5/lib/libclang.so.1', | |
611 | '/usr/lib/libclang.so', | |
612 | '/Library/Developer/CommandLineTools/usr/lib/libclang.dylib' | |
613 | ] | |
614 | libclang_found = False | |
615 | ||
616 | for lib in libclang_locations: | |
617 | if os.path.isfile(lib): | |
618 | clang.cindex.Config.set_library_file(lib) | |
619 | libclang_found = True | |
620 | break | |
621 | ||
622 | if not libclang_found: | |
62671ba0 | 623 | logging.fatal('Cannot find libclang') |
a4057fc9 | 624 | return 1 |
625 | ||
626 | # Loop over all files | |
62671ba0 | 627 | for fn in args: |
a4057fc9 | 628 | |
533918c9 | 629 | logging.info('Input file: %s' % Colt(fn).magenta()) |
a4057fc9 | 630 | index = clang.cindex.Index.create() |
631 | translation_unit = index.parse(fn, args=['-x', 'c++']) | |
10f6b9f8 | 632 | |
633 | comments = [] | |
3bd2e2f0 | 634 | traverse_ast( translation_unit.cursor, fn, comments ) |
10f6b9f8 | 635 | for c in comments: |
3018d1db | 636 | |
637 | logging.debug("Comment found for entity %s:" % Colt(c.func).magenta()) | |
638 | ||
639 | if isinstance(c, MemberComment): | |
640 | ||
641 | if c.is_transient: | |
642 | transient_text = Colt('transient ').yellow() | |
643 | else: | |
644 | transient_text = '' | |
645 | ||
646 | if c.array_size is not None: | |
647 | array_text = Colt('arraysize=%s ' % c.array_size).yellow() | |
648 | else: | |
649 | array_text = '' | |
650 | ||
72604c43 | 651 | logging.debug( |
3018d1db | 652 | "%s %s%s{%s}" % ( \ |
653 | Colt("[%d,%d]" % (c.first_line, c.first_col)).green(), | |
654 | transient_text, | |
655 | array_text, | |
656 | Colt(c.lines[0]).cyan() | |
657 | )) | |
658 | ||
3b095296 | 659 | elif isinstance(c, RemoveComment): |
660 | ||
661 | logging.debug( Colt('[%d,%d]' % (c.first_line, c.last_line)).green() ) | |
662 | ||
3018d1db | 663 | else: |
664 | for l in c.lines: | |
665 | logging.debug( | |
666 | Colt("[%d,%d:%d,%d] " % (c.first_line, c.first_col, c.last_line, c.last_col)).green() + | |
667 | "{%s}" % Colt(l).cyan() | |
668 | ) | |
a4057fc9 | 669 | |
72604c43 | 670 | try: |
671 | ||
43ee56ee | 672 | if output_on_stdout: |
673 | with open(fn, 'r') as fhin: | |
674 | rewrite_comments( fhin, sys.stdout, comments ) | |
675 | else: | |
676 | fn_back = fn + '.thtml2doxy_backup' | |
677 | os.rename( fn, fn_back ) | |
72604c43 | 678 | |
43ee56ee | 679 | with open(fn_back, 'r') as fhin, open(fn, 'w') as fhout: |
680 | rewrite_comments( fhin, fhout, comments ) | |
72604c43 | 681 | |
43ee56ee | 682 | os.remove( fn_back ) |
683 | logging.info("File %s converted to Doxygen: check differences before committing!" % Colt(fn).magenta()) | |
72604c43 | 684 | except (IOError,OSError) as e: |
685 | logging.error('File operation failed: %s' % e) | |
686 | ||
a4057fc9 | 687 | return 0 |
688 | ||
689 | ||
690 | if __name__ == '__main__': | |
62671ba0 | 691 | sys.exit( main( sys.argv[1:] ) ) |